I did quantization on a model and then the model size reduced 50% .it was one hidden layer model. its weird because when I just increase the number of neurons in its hidden layer the model reduction was greater than before (it reduced 80%).what can be the reason?
this is related to the percentage of the weight size on that layer over the sums of the weight size of the entire model I think, when you increase the number of neurons in the hidden layer, the percentage increases so the reduction is more significant.